Smart Search

Owned by Cooper Labenske
Apr 04, 2022
2 min read

Vose introducing Artificial Intelligence Capabilities

With the release of Vose 1.0, Voyager is introducing a set of AI solutions designed to help users find what they need. The functionality delivered out of the box in all of these cases, can and should be tuned to your specific content and the data that you have indexed. Models used were created based on Wikipedia, to demonstrate what is possible. Please reach out to the Support or a Sales teams to discuss tuning Voyager's smart search capabilities to your needs.

Smart Search Features

Vose includes four capabilities that rely upon machine learning models to help users find the content they need:

  • Did you mean? - If users misspells an entry, Voyager will suggest the correct word so that you never miss a thing.

  • Query Expansion - When users input a term in the keyword box, a model built on generic Wikipedia data will suggest additional similar terms and search for documents that contain any of the terms. For example, if the user enters "road," the query will be expanded to include words like "street", "drive", and "avenue". This feature can be turned to work on your data so that if you are searching for, say, a well name, you can find documents that reference that name exactly, but also any nicknames that might exist for the well.  This feature is helpful in making sure that no documents are missed in a search query.

  • Find Similar - Suggests documents that contain similar words and terms to those in the selected document. This feature is useful when looking for different versions of a document or subjects that are related.  

  • Entity Extraction  - Uses NLP (Natural Language Processing) to pull out key terms that can be used to group related documents into a taxonomies or set of filters. Out of the box, the entity extraction model can extract a set of common proper nouns — people's names, company names, and place names among others. But this module can also be turned to look for locations that are specific to your organization. For example, it can be tuned to look for a regional office name or the name for a specific product.